Accurate assembly of transcripts through phase-preserving graph decomposition.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Accurate reconstruction of full-length transcripts from RNA-sequencing (RNA-seq) data is crucial for understanding gene expression and function.
• Existing transcript assemblers face challenges in reconstructing complex transcript structures, particularly multi-exon and lowly expressed transcripts.

Purpose of the Study:

• To introduce Scallop, a novel reference-based transcript assembler designed to enhance the accuracy of transcript reconstruction.
• To improve the identification and assembly of multi-exon and lowly expressed transcripts.

Main Methods:

• Scallop employs a reference-based approach, preserving long-range phasing paths from sequencing reads.
• The assembler generates a parsimonious set of transcripts while minimizing coverage deviation.

Main Results:

• Scallop significantly outperforms StringTie and TransComb in reconstructing multi-exon transcripts by 34.5% and 36.3%, respectively.
• Scallop identifies 67.5% and 52.3% more lowly expressed transcripts compared to StringTie and TransComb.
• The assembler demonstrates higher sensitivity and precision across various coverage thresholds.

Conclusions:

• Scallop represents an advancement in transcript assembly, offering improved accuracy for complex transcript structures.
• The tool enhances the detection of lowly expressed transcripts, providing a more comprehensive view of the transcriptome.
• Scallop's performance suggests its utility for a wide range of RNA-seq analysis applications.